循证医学-病因学研究和循证医学实践

1、Evidence-basedEtiology/Harm病因研究与循证医学实践,学习目标,掌握评价病因性研究真实性原则(Validity)掌握评价病因性研究重要性原则(Importance)学会应用病因性研究证据的结果，解决临床问题(Applying),病因性研究基本知识,病因性研究基本概念与病因相关的临床问题病因性研究的主要方法病因/不良反应研究证据的分级病因性研究常用统计学指标,病因性研究基本概念（1）,病因是指引起人体发生疾病的原因。病因学是指研究疾病病因的科学。病因：致病因素（直接、间接、危险因素）研究内容：用流行病学方法研究并验证危险因素是否与疾病发生有因果关系，且评估因果联系的强弱。

2、例“吸烟与肺癌关系”,病因性研究基本概念（2）,不良反应的研究实质上也是病因学研究“因”：造成不良反应的各种因素，如各种治疗措施（药物，手术）医疗过程中临床医师经常需要考虑某种危险因素或治疗措施是否对患者有害。利是否大于弊？用他人的研究结果来回答提出的问题真实性重要性实用性,与病因相关的临床问题,该疾病是什么原因造成的？该药物或治疗措施会导致什么不良反应吗？是否需要停药？DoesexposuretoaluminumcauseAlzheimersdementia?Dostatinscausecancer?,病因性研究的主要方法,病因性研究常用统计学指标,因果相关性强度的指标RR(前瞻性)RCT,

3、cohortstudyOR(回顾性）case-controlstudyNNH(numberneededtoharm)clinicalimportance暴露多少研究对象可导致1例发病（队列研究）发生1例不良反应所需治疗的病例数（临床研究）,因果相关性强度的指标,当所研究疾病的发病率较低时，OR近似于RR，故在回顾性研究中可用OR估计RR,其解释与RR同，易于统计分析RR或OR愈高，则因果关系强度愈强RR或OR有多大才有意义，无一定的标准1.2-1.5:弱联系1.6-2.9:中等联系3.0:强联系,可信区间ConfidenceInterval,因果关系的强度外，评价精确度按一定的概率去估计总体参

4、数所在的范围95的可信区间循证医学估计总体参数假设检验：RR,有关指标的计算,1.OddsRatio,2.RelativeRisk,3.RiskReduction/Increase,4.NumberNeededtoTreat/Harm,證據的強度,TheConfusionMatrix,Alsoknownasthe2x2table,EventRate,EER=A/(A+B)试验组事件发生率CER=C/(C+D)对照组事件发生率,RRandOR,RR=EER/CER相对危险度OR=AD/BC比值比,RelativeRiskReduction,RRR=(CER-EER)/CER=1RR相对危险度减少

5、率,(Absolute)RiskReduction,ARR=CER-EER绝对危险度减少率,NumberNeededtoTreat,NNT=1/ARR得到1例有利结果需要防治的病例数,举例：ActivatedProteinCforSevereSepsis,APC=ActivatedProteinCEfficacyandsafetyofrecombinanthumanactivatedproteinCforseveresepsis.NEnglJMed.2001Mar8;344(10):699-709,EventRatesandOdds,EER=A/(A+B)=30/850=0.035CER=C/

6、(C+D)=17/840=0.020EEO=A/B=30/820=0.037CEO=C/D=17/823=0.021,OR=EEO/CEO=0.037/0.021=1.77RR=EER/CER=0.035/0.020=1.744,RRI=(EERCER)/CER=0.015/0.020=0.744=74%ARI=EERCER=0.0350.020=0.015NNH=1/ARI=66,Risk-BenefitRatio,NNT=1/ARR=1/0.06=16(治疗16个获益1个：存活)反映有利结果（越小越好）NNH=1/ARI=1/0.015=66(治疗66个损害1个：严重出血)反映不良反应（越

7、大越好）Risk-BenefitRatio=NNT/NNH=16/66=1/4,怎样解决临床问题？Howtosolveaclinicalproblem?,临床病案（ClinicalScenario）,84岁的男性，近期记忆力明显下降.高血压病，高胆固醇血症。右眼白内障术后2天，出现易激、谵妄和性格改变。无感染，贫血及代谢异常的临床证据。心理卫生中心会诊：抗精神病药物氟哌啶醇,haloperidol,奋乃静perphenazine,奥氮平,olanzapine,临床问题（InitialQuestion）,老年患者中，用传统性抗精神病药物（如氟哌啶醇,haloperidol,奋乃静perphena

8、zine,）是否会增加死亡风险性？非典型性抗精神病药物(如奥氮平,olanzapine,）是否对老年人更安全？,第一步提出问题(AskClinicalQuestions),Initialquestion:Framingtheinitialquestion:answerablePatients(population)Intervention/exposureComparisonOutcomePICO,转变成可以回答的临床问题Framingthequestion,患者类型(P)elderlypatients干预措施(I)haloperidolorperphenazine对照措施(C)olanzap

9、ine临床结局(O)death,第二步查询证据(AcquireEvidence),PICO:keywordsTypeofquestion：harm-BestevidenceLevelsofevidence-OptimalsourceofevidenceSearchingworthwhile?,病因/不良反应研究常用数据库,BestEvidence（ACPjournalclub,EBM)UptoDateMedlinePubMed:clinicalquery-etiologySumsearchOvid循证医学数据库(多库同时检索)ACPjournalclub,CochraneLibrary(CDS

10、R,CCTR,DARE),Medline,EMBASE,系统评价资料库(CochraneDatabaseofSystematicReview,CDSR)疗效评价文摘库(DatabaseofAbstractsofReviewsofEffectiveness,DARE)临床对照试验注册资料库(CochraneControlledTrialsRegister,CCTR)方法学数据库(CochraneMethodologyDatabase),检索方法,选择数据库：ACPjournalclub（oviddatabase,bestevidence）在search中，键入关键词olanzapineetiol

11、ogy（病因学）检索结果：1篇文献（摘要）找到全文,筛选结果,ACPjournalClubsummary:ConventionalantipsychoticdrugsincreasedriskfordeathmorethandidatypicalantipsychoticdrugsinelderlypatientsACPJournalClub.2007;147:23.SchneeweissS,SetoguchiS,BrookhartA,DormuthC,WangPS.Riskofdeathassociatedwiththeuseofconventionalversusatypicalantip

12、sychoticdrugsamongelderlypatients.CMAJ.2007;176:627-32,研究详情,Background:Publichealthadvisorieshavewarnedthattheuseofatypicalantipsychoticmedicationsincreasestheriskofdeathamongelderlypatients.Weassessedtheshort-termmortalityinapopulation-basedcohortofelderlypeopleinBritishColumbiawhowereprescribedcon

13、ventionalandatypicalantipsychoticmedications.Methods:WeusedlinkedhealthcareutilizationdataofallBCresidentstoidentifyacohortofpeopleaged65yearsandolderwhobegantakingantipsychoticmedicationsbetweenJanuary1996andDecember2004andwerefreeofcancer.Wecomparedthe180-dayall-causemortalitybetweenresidentstakin

14、gconventionalantipsychoticmedicationsandthosetakingatypicalantipsychoticmedications.,Results:,Of37241elderlypeopleinthestudycohort,12882wereprescribedaconventionalantipsychoticmedicationand24359anatypicalformulation.Withinthefirst180daysofuse,1822patients(14.1%)intheconventionaldruggroupdied,compare

15、dwith2337(9.6%)intheatypicaldruggroup(mortalityratio1.47,95%confidenceintervalCI1.391.56).Multivariableadjustmentresultedina180-daymortalityratioof1.32(1.231.42).Incomparisonwithrisperidone（利培酮）,haloperidol（氟哌啶醇）wasassociatedwiththegreatestincreaseinmortality(mortalityratio2.14,95%CI1.862.45)andloxa

16、pine（洛沙平）thelowest(mortalityratio1.29,95%CI1.191.40).Thegreatestincreaseinmortalityoccurredamongpeopletakinghigher(abovemedian)dosesofconventionalantipsychoticmedications(mortalityratio1.67,95%CI1.501.86)andduringthefirst40daysafterthestartofdrugtherapy(mortalityratio1.60,95%CI1.421.80).Resultswerec

17、onfirmedinpropensityscoreanalysesandinstrumentalvariableestimation,minimizingresidualconfounding.,结论,Interpretation:Amongelderlypatients,theriskofdeathassociatedwithconventionalantipsychoticmedicationsiscomparabletoandpossiblygreaterthantheriskofdeathassociatedwithatypicalantipsychoticmedications.Un

18、tilfurtherevidenceisavailable,physiciansshouldconsiderallantipsychoticmedicationstobeequallyriskyinelderlypatients.,第三步评价证据AppraiseEvidence,证据的真实性Aretheresultsvalid?证据的重要性Whataretheresults?,证据的真实性Aretheresultsvalid?,1研究方法的论证强度TypeofReportsonEtiology/Harm,哪种研究方法？论证强度如何？是否源于真正的人体试验？Werethereclearlydef

19、inedgroupsofpatients,similarinallimportantwaysotherthanexposuretothetreatmentorothercause?,本研究,Objective:Inelderlypatients,associationofconventionaloratypicalantipsychoticdrugs(APDs)withdeath?Design:CohortstudyParticipants:37241patients65yofageoralconventional(n=12882,meanage80y)atypical(n=24359,mea

20、nage80y).Exclusioncriteria:canceranduseofAPDsintheyearbeforetheindexdate.,2两组结局暴露因素的测量方法是否一致？,Weretreatments/exposuresandclinicaloutcomesmeasuredinthesamewaysinbothgroups?(Wastheassessmentofoutcomeseitherobjectiveorblindedtoexposure?)Weretheoutcomesandexposuresmeasuredinthesamewayinthegroupsbeingcom

21、pared?,CohortStudy,Surveillancebias:监测偏倚偏倚的控制客观指标（Objectiveoutcome）：病死率主观指标（Subjectiveoutcome）:Blinding举例：乙型肝炎与肝癌关系的研究,3.随访时间及失访率,Wasthefollow-upofthestudypatientssufficientlylong(fortheoutcometooccur)andcomplete?举例：HP与胃癌：5年（无差异），10年（显著差异）失访超过20？-结果将失去真实性,4病因/不良反应研究结果是否符合病因诊断原则,Dotheresultsoftheharm

22、studysatisfysomeofthediagnostictestsforcausation?,Isitclearthattheexposureprecededtheonsetoftheoutcome?因果效应的先后顺序仅见于前瞻性研究Isthereadoseresponsegradient?因果效应的相关程度，剂量依赖（吸烟与肺癌）Isthereanypositiveevidencefroma“dechallengerechallenge”study?符合流行病学规律-危险因素减弱，发病减少,Istheassociationconsistentfromstudytostudy?不同研究，

23、结果一致（HP与胃癌）Doestheassociationmakebiologicalsense?充分的生物学依据（CCB与癌症，坏血病与水果蔬菜）,KeyPoints,1.Werethereclearlydefinedgroupsofpatients,similarinallimportantwaysotherthanexposuretothetreatmentorothercause?研究方法的论证强度2.Weretreatments/exposuresandclinicaloutcomesmeasuredinthesamewaysinbothgroups?测量方法一致3.Wasthefo

24、llow-upofthestudypatientssufficientlylong(fortheoutcometooccur)andcomplete?随访时间及失访率,证据的重要性Whataretheresults?,1.因果联系强度,Whatisthemagnitudeoftheassociationbetweentheexposureandoutcome?Howstrongistheassociationbetweenexposureandoutcome?RRORNNH,2.结果是否准确？,Whatistheprecisionoftheestimateoftheassociationbet

25、weentheexposureandoutcome?Howpreciseistheestimateofrisk?95%CI,ConventionalAPDvsAtypicalAPDAssociationwithdeath,第四步应用证据HowcanIapplytheresultstomypatient?,病情相似,Isourpatientsodifferentfromthoseincludedinthestudythatitsresultscannotapply?Werethestudypatientssimilartomypatient?基于纳入和排除标准,本研究,Patients:65yo

26、fage,60-65%womenUsed1medicalservice,andfilled1prescriptioninthetwo6-monthintervalsbeforetheindexdate.Exclusioncriteria:canceranduseofAPDsintheyearbeforetheindexdate.AtypicalAPDs:risperidone,quetiapine,olanzapine,andclozapineConventionalAPDs:loxapine,haloperidol,chlorpromazine,trifluoperazine,thioridazine,pimozide,promazine,p